Hello there,
My converter uses multiple T-legs in parallel and I wish to represent that in PLECS. I created a detailed model of one leg and tried to scale it using a circuit arrangement shown in the figure below.
I then built a circuit to compare the performance of the original block and the scaled one with N=1 and I learned that the scaled one is inaccurate at higher frequencies. I tried configuring V and I sources to “Non-causal zero-order hold” and to “Zero-order hold” and found no difference in performance. I had “Allow state-space inlining” turned on in all experiments.
It would be really convenient if I could represent parallel legs by setting the N value in the mask, not having to instantiate a group every time I need to analyze a different converter configuration. Is there a way to achieve this?
Jovan, could you please post a representative model that shows your implementation and scenario where the results don’t align for analysis / discussion?
I tried to reproduce the issue this morning and could not, leading me to think I had something wrong in the library.
Please review the way I set up the library - my scaled AMT legs are referencing the single one using the model reference because they reside in the same file. Also, please advise if I should use zero order hold or non causal zero order hold in the coupled sources.
Jovan, Thanks for posting these files. As you suggested, your scaling logic seems to be working quite well. I tested up to N=3. I hope other users find this approach useful!
Also, please advise if I should use zero order hold or non causal zero order hold in the coupled sources.
The discretization behavior setting will not have an impact in your model because you enabled state-space inlining setting and are not using a discrete solver.
